Abstract. The HIV epidemic, the increasing prevalence of diabetes in China and India, and the spread of drug- resistant Mycobacterium tuberculosis (Mtb) are perpetuating the global tuberculosis (TB) epidemic. While BCG is widely used as a vaccine, it efficacy in preventing pulmonary TB is limited. Therefore, TB vaccine development continues to be an important priority. Several vaccine candidates are being evaluated clinically, but since their ability to prevent TB must be tested empirically, progress is slow and expensive. While there is general agreement that T cells mediate immunity to Mtb, there is little ability to predict which antigens are targets of protective immune responses. For a vaccine to be effective, the elicited antigen-specific T cells must recognize infected cells. However, there may be an important difference between the repertoire of antigens presented by DC in the LN during T cell priming occurs, and the antigens that are presented by infected macrophages in the lung. Such differences in antigen processing arise for many reasons: 1) the DC that prime T cells are not always infected but have acquired nonviable Mtb antigens; 2) DC and macrophages present antigen differently; 3) inflammatory signals alter antigen presentation (e.g., IFN?) in the lung. Whether an infected cell presents bacterial epitopes at a sufficient density for T cell recognition may depend on the number, location, and metabolic activity of intracellular bacteria, which can alter bacterial antigen production. Finally, the idea that Mtb evades detection of the immune response is well-documented and the mechanisms are being elucidated. In a limited number studies, immunodominant T cell responses fail to recognize infected cells. One way to think about this phenomenon is that uninfected APC acquire and present Mtb proteins, which both elicits and perpetuates T cell responses to abundant antigens; however, these antigens are inefficiently presented by Mtb-infected cells. We propose to develop a vaccine that preferentially stimulates T cells that recognize Mtb infected cells. As part of this proposal, we will develop an assay to quantify vaccine-elicited T cells that recognize Mtb infected cells, which we envision as being essential for pre-clinical and clinical vaccine development. By vaccinating mice with macrophages or DC that have been infected in vitro with the 18b auxotroph of Mtb, we will test whether we can skew the immune response toward antigens that are presented by Mtb-infected macrophages. This will allow us to test the concept that a vaccine that elicits T cells that recognizes Mtb-infected cells will have greater efficacy in protecting the host. If successful, identification of which antigens are targets of protective immunity would greatly accelerate vaccine development. Our overarching model is that antigen presentation by uninfected DC determines the hierarchy of Mtb-specific T cells elicited during infection. However, the relative abundance of Mtb epitopes presented by infected macrophages determines the hierarchy of protective T cells. ! !